Intelligent Character Recognition
Automatically recognizing characters are part of OCR applications. There has been so many different approaches to this problem including mechanical, vibrational as well as optical methods. Although modern OCR algorithms are succesful in capturing individual characters and then combining these characters into groups to form words and then sentences, they are prone to error when it comes to difficult light settings, fonts and many other reasons.
Most important drawback of standard OCR algorithms is that the algorithm does not have semantic capabilities as such detected characters can not directly be used to extract the meaning of written symbols.
However, most business apllications require
extracting information such as financial, customer related etc.
OCR algorithms augmented with Machine Learning offers the best approach for information processing and business analytics, thereby providing better business automation opportunities..
Our team implemented Machine Learning and Deep Learning algorithms to tackle with this problem for
Business Document Automation , especially extracting financial information, but also gathering information about companies
by enabling calculations such as Altman's z-score regarding the financial health and investibility of a company.
The outcome is not only limited with direct financial information. Using Natural Language Processing our team was able to extract information on key decision makers and their demographic information from any document (word, pdf, or scanned documents).